Job Automation: Deliverance or Dystopia?
Automation, Machine Learning (ML) and Artificial Intelligence (AI) are no longer “future” concepts, but do their arrival announce humanity’s liberation from the ancestral burden of work or a social crisis from unemployment and inequality? Technological advancement has always been a driver of social progress and economic growth. Nevertheless, the so-called Fourth Industrial Revolution seems quite different from the previous “revolutions” (Bloem et al.,2014), given that machines are surpassing humans, not only in strength and speed, but also in cognitive abilities. The exponential advancement in AI, ML, Robotics, Automation and even Quantum Computing is transforming society more quickly than we have ever experienced. Most of the advancements thus far have been in known as “weak AI”, meaning AI conceived as a tool for solving problems, whereas “strong AI” is the generation of an actual human-level “intelligence” (Flowers, 2019), and is the next major step researchers are competing to reach in technology. Once strong AI is achieved, technological singularity may be an unavoidable reality (Müller and Bostrom, 2016) and potentially exceed the cognitive skills of humans, meaning that if human intelligence is not modified, we would stop being the most intelligent beings on earth. Experts believe that superintelligence may be achieved “within a generation” (Hornigold, 2018).
Automation: The Rewards and the Risks
All this raises questions from many domains such as ethics, the environment, and particularly the future of work and its consequences for inequality. Job automation is the execution by a machine of a function previously done by a human (Parasuraman and Riley, 1997). It has freed humans from time-consuming and repetitive physical and/or cognitive tasks, allowing for us to focus on more complex, creative, and emotional tasks. With the development of ML, great advances have been achieved, including in transportation, drones, education, manufacturing, human resources, cybersecurity, home, healthcare, finance, hazardous environment, and more. These advances have positive externalities, given that if AI does most of the work and produces more, goods and services will be plentiful. For example, some estimations show that the AI health market growth can potentially create US$150 billion in annual savings for the United States healthcare economy by 2026 (Siau and Wang, 2019).
However, significant risks also exist, including that the inner working of these self-learning machines is a “black box”, which makes it difficult to trust them. Also, Machine Learning using Big Data can be susceptible to human biases (Lewis and Monett, 2017). Another concern is that AI could lead to some global catastrophe or even human extinction (Bostrom, 2014), but this is still an under-researched area (Siau and Wang, 2019). Furthermore, AI may create unemployment and widen the inequality of wealth (Siau and Wang, 2018), as AI enables companies to cut down on human labor, which so far has been mostly concentrated on jobs requiring routine, repetitive, and predictable tasks. This way, owners of AI-driven companies increase their wealth, while the unemployed lose their source of income. One statistic that highlights this troubling reality is that, in 2014, while the three largest companies in Detroit generated the same revenue than the three largest companies in Silicon Valley, the latter had 10 times fewer employees (Bossmann, 2016).
Research literature has noted that wages and educational level are negatively correlated with automation risk, and jobs that are based in creativity, social intelligence, strategy, and empathy have the lowest automation risk (Frey and Osborne, 2017). Acemoglu and Autor (2011), as well as Acemoglu and Restrepo (2018), talk about Enabling Technologies, which expand the productivity of labor and lead to higher employment and wages, and Replacing Technologies, which substitute labor, making workers less useful and lowering their wages. In this scenario, those well prepared with complementary skills would benefit the most from technological innovation. However, even if the utopian post-work society may not be realized, the rapid advancement of AI suggests that humans will need to figure out how contribute to society and to find meaning in non-labor activities.
Regional Variations and the Digital Divide
Deloitte, in partnership with Oxford University, propose that 35% of jobs will be at risk of disappearing during the next 20 years (Wakefield, 2016). Moreover, estimations from Oxford University show that at least 47% of U.S. jobs and 54% of those in Europe are at a high-risk of automation (Bregman, 2017). Historian Yuval Noah Harari (2016) claims that AI will create a ”useless class” of humans, who will be not only unemployed but also unemployable. All this, of course, if public policy doesn’t step in.
It is important to note that, currently, the workforce displaced by growing automation in the manufacturing sector seems to be absorbed by the services sector, which represents almost 74% of the workforce in developed economies and 52% of the global workforce (ILO, 2016). The same trend occurred when job losses in the agricultural sector were substituted by the manufacturing sector during the First Industrial Revolution. Furthermore, there is the paradox that this overabundance of labor slows the progress of automation, given that the automatable work may be done more cheaply by humans than by the investment needed to replace them with machines (Smith, 2020). In that sense, evidence from Latin America, shows that given different occupational structures, the exposure to automation is heterogeneous across demographic and socioeconomic groups and countries, with the skills gap being the most notable factor (Gasparini et al., 2021).
In this context, the Digital Divide is determinative of economic performance of individuals. Being digitally literate is imperative for accessing economic, social and political opportunities, including citizen’s rights. In this way, the digital divide can lead to a further increase of existing inequalities and the continuous social exclusion of marginalized groups. It is worth noting that just about 60% of the world’s population is able to access and use the internet, according to The World Bank. Even if the digital transformation offers unprecedented opportunities for inclusive and sustainable economic development, access to digital education that would lead to better jobs in a world where programming and human-machine interaction skills will become more and more important, might be a challenge to close this divide without policy measures being taken.
The Future of Inequality and Redistribution
This brings up the same debates about wealth distribution that have existed for centuries, but in a new light, as the inequality scenario could be even more extreme. From Universal Basic Income (UBI) to taxation of robots, many ideas have been proposed to address these issues. The feasibility of UBI is uncertain and there are reasons to think it might generate other problems, such as mental health issues (Siau and Wang, 2019) or discouragement from seeking retraining and reemployment, as well as causing those still paying high taxes to flee the countries that implement UBI (Re-educating Rita, 2019). So policies to facilitate labor market flexibility and mobility, as well as education and retraining appear to be the most straightforward solutions (Clifford, 2018), assuming that new jobs or human-robot cooperative opportunities will be created. It has been theorized that 65% of children will have jobs that do not exist yet (Kasriel, 2018). This could imply the necessity of a whole new education structure for misplaced people being able to remain competitive in the job market through lifelong learning opportunities (Card et al., 2018; OECD, 2017). For this, the financial benefits reaped from automation should be used to fund continuous education. Therefore, an important challenge for policymakers and educational institutions is the identification of complementary skills for future work, emphasizing scientific and mathematical abilities, as well as soft skills such as communication, flexibility, creativity, and adaptability.
But in the long run, there will be a need for broader redistribution policies. In Anthony Atkinson’s ‘Inequality, What can be done?’ (2015), the first proposal refers to the direction of technological change, calling for policymakers to encourage innovation which increases the employability of workers and human-centered provision of goods and services, while also guaranteeing public employment. Another proposal refers to the implementation of a capital endowment (minimum inheritance) to be paid to all citizens upon reaching adulthood, so that everyone is an owner of the current and potential technologies, and the technological dividends are distributed more evenly throughout the population.
For emerging economies, particularly those in Africa and Latin America and the Caribbean, the automation process makes it harder to achieve economic growth, so inequality between countries may increase too. As mentioned, LDCs (Least Developed Countries) offer cheaper labor which enables them to avoid job automation, but automation has significantly decreased the need for companies to outsource their production to LDCs (Re-educating Rita, 2019), which may make growth more difficult. What’s worse is that these countries may need to rely on the AI systems created in developed countries, which may not be the most adequate for the conditions in developing economies. Either way, given a decreasing pattern of RTC (Routine task content) in education, and as most literature in LDCs finds, there is no evidence for polarization in the labor market as pronounced as what is found in developed economies. Maloney and Molina (2016) suggest some possible reasons for this, including different initial occupational distributions, impact of off-shored jobs and the effect of new technologies in fostering sectors that create middle-skill jobs. However, more research is needed to understand these factors. For instance, in Latin America and the Caribbean, the automation process is more likely to affect the structure of employment, with unskilled and semi-skilled workers likely to bear a disproportionate share of the adjustment costs compared to skilled workers, making automation more of a threat to income equality than overall employment (Brambilla et al., 2021).
Humanity needs to be proactive, rather than reactive, in managing the consequences of technological advancement. But are decision-makers thinking strategically about the forces of disruption and innovation shaping our futures? It seems like they are too often trapped in traditional, linear thinking, or too absorbed by the multiple crises demanding their attention to fully prepare for this massive shift that will affect most of the world’s population in some way (Chuah et al., 2018). For the rest of us, as stated by Sam Harris, the first step would be to just start thinking about these issues, given that a technological explosion seems to be unavoidable.
About the author: Evelin Lasarga is a Project and Research Officer with Data-Pop Alliance currently residing in her home country of Uruguay.
- Acemoglu, D., and Autor, D. (2011). Skills, Tasks and Technologies: Implications for Employment and Earnings. In Handbook of Labor Economics, Volume 4, edited by Orley Ashenfelter and David E. Card. Amsterdam: Elsevier.
- Acemoglu, D., and Restrepo, P. (2017). Robots and Jobs: Evidence from US Labor Markets. NBER Working Paper 23285, National Bureau of Economic Research, Cambridge, MA.
- Atkinson, A. B. (2015). Inequality: what can be done?. Cambridge, Massachusetts: Harvard University Press.
- Bloem, J., Van Doorn, M., Duivestein,S., Excoffier, D., Maas, R., and Van Ommeren, E. (2014). The Fourth Industrial Revolution —Things to Tighten the Link between IT and OT. In Sogeti VINT 2014.
- Bostrom, N. (2014). Superintelligence: Paths, dangers, strategies. Oxford University Press.
- Bossmann, J. (2016). Top 9 ethical issues in artificial intelligence. World Economic Forum.
- Brambilla, I., César, A., Falcone, G., Gasparini, L. and Lombardo, C. (2021). The Risk of Automation in Latin America. Documentos de Trabajo del CEDLAS No 281, Junio, 2021, CEDLAS-Universidad Nacional de La Plata.
- Bregman, R. (2017). A growing number of people think their job is useless. Time to rethink the meaning of work. World Economic Forum.
- Card, D., Kluve, J., and Weber, A. (2018). What Works? A Meta-Analysis of Recent Active Labor Market Program Evaluations. Journal of the European Economic Association 16 (3): 894–931.
- Chuah, L. L., Loayza, N., and Schmillen, A. D. (2018). The Future of Work: Race with—not against—the Machine. Available at SSRN 3400509.
- Flowers, J. C. (2019). Strong and Weak AI: Deweyan Considerations. In AAAI Spring Symposium: Towards Conscious AI Systems (Vol. 22877).
- Frey, C. B., and Osborne, M. A. (2017). The future of employment: How susceptible are jobs to computerization? Technological Forecasting and Social Change, 114, 254–280.
- Gasparini, L., Lombardo, C., Brambilla, I., César, A. and Falcone, G. (2021). Routinization and Employment: Evidence for Latin America. Documentos de Trabajo del CEDLAS No 276, Marzo, 2021, CEDLAS-Universidad Nacional de La Plata.
- Harari, Y. N. (2016). Homo deus: A brief history of tomorrow. Harvill Secker.
- Hornigold, T. (2018). When Will We Finally Achieve True Artificial Intelligence. Singularity Hub. Retrieved from https://singularityhub.com/2018/01/01/when-will-we-finally-achieve-true-artificial-intelligence/#sm.000 01azv7jc2gld26qgto4a94uxnr
- Kasriel, S. (2018). The future of work won’t be about college degrees, it will be about job skills. CNBC. Retrieved from https://www.cnbc.com/2018/10/31/the-future-of-work-wont-be-about-degrees-it-will-be-about-skills.html
- Lewis, C., and Monett, D. (2017). AI & Machine Learning Black Boxes: The Need for Transparency and Accountability. KDnuggets. Retrieved from https://www.kdnuggets.com/2017/04/ai-machine-learning-black-boxes-transparency-accountability.html
- Maloney, W. and Molina, C. (2016). Are automation and trade polarizing developing country labor markets, too? The World Bank.
- Müller, V. C., and Bostrom, N. (2016). Future progress in artificial intelligence: A survey of expert opinion. In Fundamental issues of artificial intelligence (pp.553–570). Springer International Publishing.
- Parasuraman, R., and Riley, V. (1997). Humans and automation: Use, misuse, disuse, abuse. Human Factors, 39(2), 230–253.
- Re-educating Rita. (2019). Artificial intelligence will have implications for policymakers in education, welfare and geopolitics. The Economist. Retrieved from https://www.economist.com/special-report/2016/06/25/ re-educating-rita
- Siau,K., and Wang, W. (2018). Building Trust in Artificial Intelligence, Machine Learning, and Robotics. Cutter Business Technology Journal, 31(2), 47–53.
- Siau, K. and Wang, W. (2019). Artificial intelligence, machine learning, automation, robotics, future of work and future of humanity: A review and research agenda. Journal of Database Management (JDM), 30(1), 61-79.
- Smith, J. E. (2020). Smart Machines and Service Work. Automation in an Age of Stagnation, London, Reaktion Books.
- Wakefield, J. (2016). Foxconn replaces 60,000 factory workers with robots. BBC. Retrieved from http://www. bbc.com/news/technology-36376966?SThisFB